Shared definitions cut ad settlement close from seven days to three

Separating master data, metadata and transactions, then versioning calculation rules, helped one operation reduce settlement disputes by 40% and improve receivables accuracy to 99.5%.

In many advertising and e-commerce operations, the most time-consuming monthly argument is not about what happened in the business, but about which figure should count as the truth. One team may rely on the media platform, another on the finance system, and a third on the business middle platform, only to find that the same settlement sheet still produces different answers. The result is familiar: repeated reconciliations, delayed closes and disputes that consume energy without resolving the underlying problem.

The core issue is often not a faulty calculation, but a lack of shared meaning. As AlouData argues in its discussion of semantic governance, unifying data alone does not automatically create a common business understanding. That distinction matters in sectors such as advertising, where the same label can conceal different sources, timings and calculation rules. FineBI has made a similar point in its guidance on standardising metrics, noting that inconsistency usually begins when departments define the same indicator in different ways and then assume they are talking about the same thing.

That is why it helps to separate three levels that are often confused in practice: master data, metadata and transactional data. Master data covers the core entities that must be recognised consistently across systems, such as customers, media, accounts and organisational units. Metadata and data standards define what each field means, where it comes from, how it should be measured and which rules apply. Transactional data records what actually happened on a given day, whether that is media spend, top-ups, refunds or settlements. If those layers are not kept distinct, a settlement can inherit errors long before it reaches the finance team.

The consequences spread quickly. A field with the same name may carry a different commercial meaning from one media source to another. A value taken from the source system may not be the same as a value suitable for settlement. Data pulled through APIs or RPA can also be late, incomplete or duplicated if the process is not checked properly. SmartBI’s explanation of multi-source metric governance stresses the need for mapping rules, auditing and structured governance precisely because field differences do not disappear simply because a platform can ingest them.

A workable response begins with setting clear boundaries and responsibilities. FineBI and SmartBI both emphasise the importance of cross-department cooperation, and that is borne out in practice: finance, operations and product teams need to agree on definitions, sources and version control, rather than relying on informal understandings. In the case described here, the team first identified which business objects had to be governed centrally, then wrote a shared data dictionary, recorded the original value and the calculated value separately, and built versioned rules so historical settlements could be traced back to the exact logic in force at the time.

The architecture that followed was less a single interface than a traceable chain. Media data was collected through APIs or RPA, stored in its original form, cleaned and validated against the dictionary, then mapped into standardised values before being used by the business platform for settlement and BI reporting. Automated comparison and anomaly alerts made it possible to catch missing data, timing issues and value discrepancies earlier in the cycle, while daily reconciliation helped test whether the rules actually worked under live conditions. Over two months, that approach helped cut the monthly closing cycle from seven days to two or three and reduced disputes by roughly 40%, while improving receivables accuracy to 99.5%.

The larger lesson is that data governance only pays off when it resolves semantic disagreement as well as technical inconsistency. A shared dictionary, a stable identifier and a versioned rule set do more than tidy up a database: they give operations and finance the same language for deciding what counts, where it came from and how it should be settled. Once that language is common, month-end stops being a debate about whose number to trust and becomes a process of checking a single, agreed version of the business.

Source Reference Map

Inspired by headline at: ^[1]

Sources by paragraph:

Source: Fuse Wire Services